Semiconductor inventories are falling but still above historical averages, with AI and memory chains offering structural opportunities

Overview

This report uses a semiconductor supply-chain inventory tracking framework to assess inventory and days-on-hand changes across customers, distributors, and manufacturers. The core conclusion is that customer and distributor inventories continue to decline and in some areas are below historical averages, but manufacturer inventories remain elevated, leaving total supply-chain inventories still above historical average. The report argues that absent a major supply shock or further demand-supply tightening, inventories may begin to be replenished gradually from this quarter onward, but macro uncertainty limits companies' willingness for large-scale restocking, with automotive parts-related companies being especially cautious.

Core views

The report takes a structurally constructive view on semiconductors. On one hand, ample supply, shorter lead times, and uncertain end demand from autos and other sectors make overall inventory normalization relatively slow; on the other hand, AI-related product demand for wafer capacity, broader product-price pass-through, and rising memory costs create opportunities in select subsegments and stocks. The report shows greater preference for defensive high-end semiconductor companies, AI-related compute and networking companies, the memory chain, and semiconductor capital equipment beneficiaries.

Analysis framework

The report uses inventory days, order trends, lead times, month-over-month changes, historical averages, and four-quarter rolling averages as primary metrics, splitting inventory into customers, distributors, and semiconductor manufacturers, and further examining subsegments including autos, consumer, contract manufacturing, compute mobility, storage, analog and MCU, smartphones, memory, and components.

Methodology notes

• Inventory cycle analysisSupply-chain inventory tracker

  Inventory turnover days

  Using DOI to measure the relative level of inventory at each supply-chain stage against the speed of sales or cost digestion, helping to judge inventory pressure and replenishment room.

• Historical comparison10-year median and seasonality comparison

  Deviation from historical average

  Compares current DOI with the 10-year median, normal seasonal shifts, and the past four-quarter averages to separate normal seasonal destocking from abnormal inventory pressure.

• Supply-chain segmentationThree-tier inventory framework for customers, distributors, and manufacturers

  Inventory location

  Allocates inventory pressure across customers, distribution channels, and semiconductor companies themselves to avoid judging industry supply-demand status using only total inventories.

• Demand-supply tightnessLead-time observation

  Lead time

  Uses lead-time changes for analog chips, MCU, and the broader supply chain to assess whether supply is tight or loose and to help identify price and capacity constraints.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• ADI, NXP

  Preferred defensive high-end semiconductor names

  Strengths

  Product positioning is high-end, demand resilience is relatively stronger, and they may be suitable for allocation when macro conditions are uncertain and replenishment pace is slow.

  Weaknesses

  Still affected by generally elevated supply-chain inventories and weak replenishment willingness in automotive parts.

  Comparison

  Compared with more cyclical categories, the report leans toward positioning these as defensive choices.

  Risks

  If end demand continues to weaken or auto-chain destocking is prolonged, valuations and order recovery may face pressure.

• ALGM

  AI-related opportunity name

  Strengths

  Benefits from structural demand tied to AI-related products and production-capacity reallocation.

  Weaknesses

  No specific financial forecasts or target price are provided by the report; evidence is primarily directional from industry supply-demand views.

  Comparison

  Compared with traditional inventory-cycle names, it is more dependent on continued AI-related demand.

  Risks

  A slowdown in AI demand, weaker-than-expected order realization, or changes in capacity allocation could weaken the investment thesis.

• NVDA, AVGO, ALAB

  AI beneficiaries in compute and networking

  Strengths

  Strong AI compute and networking demand, with wafer capacity shifting toward AI-related products potentially supporting tight supply-demand and pricing.

  Weaknesses

  Valuations may already reflect many optimistic assumptions and are sensitive to realization of growth.

  Comparison

  Compared with traditional semiconductor names, these are less encumbered by the current inventory cycle and more exposed to AI capex.

  Risks

  A slowdown in AI capital spending, changes in supply-chain bottlenecks, or regulatory constraints could increase volatility.

• SNDK, MU

  Memory-cycle beneficiaries

  Strengths

  The report notes that rising memory costs have not yet eased, which may improve pricing and earnings expectations.

  Weaknesses

  Rising memory DOI month-over-month indicates continued pressure at the inventory level.

  Comparison

  Compared with defensive analog names, these have greater cyclical sensitivity and are more dependent on the persistence of price increases.

  Risks

  If end demand is weak or new supply comes to market, memory price improvements may not persist.

• AMAT, KLAC, MKSI

  Semiconductor capital equipment watchlist

  Strengths

  AI-related capacity tightness and wafer-capacity reallocation may support equipment demand.

  Weaknesses

  Sector view is in-line, indicating the equipment segment is not unconditionally bullish.

  Comparison

  Compared with pure-play chip designers, the equipment chain is more sensitive to capex cycles.

  Risks

  Delayed capex by foundries, export restrictions, or weaker-than-expected sector replenishment may impact orders.

Key data

• Total supply-chain inventoryDown 5 days month-over-month, still 31 days above historical averageThe decrease is smaller than the typical seasonal drop of 11 days, indicating total inventory pressure remains elevated.
• Customer inventory52 days, down 5 days month-over-monthThe decline is greater than the normal seasonal decline of 3 days and is about 2 days below historical average.
• Distributor inventory63 days, down 2 days month-over-monthInventory turnover days are still 9 days above historical median, but trend has declined over the past four quarters.
• Semiconductor company inventory111 days, up 3 days month-over-monthHigher than historical average by 22 days; production-side inventory build remains the main contributor to elevated total inventories.
• Analog chips and MCU lead timesAbout 30% and 40% above normal levels, respectivelyAlthough industry recovery is gradual, supply cycles for certain categories still indicate tightness.
• Work-in-process inventory indexDown 3.1% month-over-month, up 2.2% cumulatively over the past four quartersShows coexistence of short-term destocking and medium-term inventory accumulation.
• Memory group DOIUp 8 days month-over-month, 9 days above historical averageUpward pressure on memory costs has not yet shown clear relief.

Impact & implications

From an investment perspective, the report does not support a broad, indiscriminate bullish view on the semiconductor inventory cycle; instead, it emphasizes differences in inventory location and segment-level supply-demand conditions. Low customer and channel inventories are supportive of future replenishment, but elevated manufacturer inventories and macro uncertainty limit overall sector resilience. Fundamentals for AI, memory, and defensive high-end semiconductors are more supportive, while automotive parts and demand-sensitive areas still need clearer evidence of order recovery.

Risks

• Manufacturer inventories remain elevated, which may delay broad sector replenishment and price normalization.
• Rising macro uncertainty, with companies lacking motivation for large-scale inventory top-up.
• Automotive parts manufacturers continue to show weak inventory and order recovery, potentially weighing on related semiconductor demand.
• AI and memory opportunities are highly sensitive to demand durability, sustained price increases, and capacity allocation.
• The report contains extensive regulatory disclosures and conflict-of-interest statements involving certain companies receiving investment-banking services or shareholding relationships, so views should be interpreted carefully with disclosures.

What to watch

• Whether customer inventories continue below historical averages and trigger genuine replenishment orders.
• Whether distributor inventory turnover days continue to decline further from 9 days above historical median.
• Whether semiconductor company DOI starts to fall from 111 days, confirming easing production-side inventory pressure.
• Whether analog chip and MCU lead times remain above normal levels, confirming demand-supply tightness.
• Whether rising memory costs and prices continue to translate to related companies' earnings.
• Whether AI-related wafer capacity allocation further crowds out non-AI product supply.
• Whether automotive OEM and automotive parts supplier order indices recover toward the median.